Available Models

There are a couple of different ways to check which models are available in your current session.

px.models.list_models()
px.models.list_providers()
px.models.list_provider_models()
px.models.get_model()

`px.models.list_models()`

It is possible to list all available models as follows:


provider_models = px.models.list_models()
 
for provider_model in provider_models:
  print(provider_model)


(claude, haiku)
(claude, opus)
(claude, sonnet)
(gemini, gemini-2.0-flash)
(gemini, gemini-2.0-flash-lite)
(gemini, gemini-2.5-pro-preview-03-25)
(openai, chatgpt-4o-latest)
(openai, gpt-4.1)
...

Returns a list of px.types.ProviderModel.

Parameters

Option	Type	Default Value	Description
`model_size`	`str`	`None`	One of `'small'`, `'medium'`, `'large'`, or `'largest'` as string or `px.types.ModelSizeType` enum. If provided, only models of this size are returned. `'largest'` is for the largest models of each provider.
`return_all`	`bool`	`False`	Previously failed models are cached by default. If you set this option to `True`, all models are returned, including both successfully available and previously failed ones.
`verbose`	`bool`	`False`	If `True`, the function will print the details of how the models are filtered.
`clear_model_cache`	`bool`	`False`	Model results (both successful and failed) are cached by default, even without a specified cache path. Setting this option to `True` clears the cache and forces a fresh check of all models. • This option is handy when API keys are changed, quotas of some providers are changed, etc.

`px.models.list_providers()`

It is possible to list all available providers as follows:


providers = px.models.list_providers()
 
for provider in providers:
  print(provider)


claude
gemini
openai
...

Returns List[str].

Parameters

Option	Type	Default Value	Description
`verbose`	`bool`	`False`	If `True`, the function will print the details of how the providers are generated.
`clear_model_cache`	`bool`	`False`	Model results (both successful and failed) are cached by default, even without a specified cache path. Setting this option to `True` clears the cache and forces a fresh check of all providers. • This option is handy when API keys are changed, quotas of some providers are changed, etc.

`px.models.list_provider_models()`

This function lists all available models for a given provider.


provider_models = px.models.list_provider_models(provider='claude')
 
for provider_model in provider_models:
  print(provider_model)


(claude, 3-haiku)
(claude, 3-sonnet)
(claude, 3.5-sonnet)
(claude, 3.5-sonnet-v2)
(claude, haiku)
(claude, opus)
(claude, sonnet)

Returns List[str].

Parameters

Option	Type	Default Value	Description
`model_size`	`str`	`None`	One of `'small'`, `'medium'`, `'large'`, or `'largest'` as string or `px.types.ModelSizeType` enum. If provided, only models of this size are returned. `'largest'` is for the largest model of the provider.
`provider`	`str`	`None`	The provider to list models for. Check available providers on ProxAI Github Repo .
`verbose`	`bool`	`False`	If `True`, the function will print the details of how the models are generated.
`clear_model_cache`	`bool`	`False`	Model results (both successful and failed) are cached by default, even without a specified cache path. Setting this option to `True` clears the cache and forces a fresh check of all models. • This option is handy when API keys are changed, quotas of some providers are changed, etc.

`px.models.get_model()`

This function returns the model details for a given provider and model name.


provider_model = px.models.get_model(provider='claude', model='3-haiku')
 
print(repr(provider_model))


ProviderModelType(provider=claude, model=3-haiku, provider_model_identifier=claude-3-haiku-20240307)

Returns px.types.ProviderModelType.

Parameters

Option	Type	Default Value	Description
`provider`	`str`	`None`	String value of the provider. Check available providers on ProxAI Github Repo .
`model`	`str`	`None`	String value of the model. Check available models for a given provider on ProxAI Github Repo .
`allow_non_working_model`	`bool`	`False`	If `True`, the function will return the model even if it is not working. If `False`, the function will raise an error if the model is not working.
`verbose`	`bool`	`False`	If `True`, the function will print the details of how the models are generated.
`clear_model_cache`	`bool`	`False`	Model results (both successful and failed) are cached by default, even without a specified cache path. Setting this option to `True` clears the cache and forces a fresh check of all models. • This option is handy when API keys are changed, quotas of some providers are changed, etc.

Details of how models are checked

The available models are determined as follows:

It checks environment variables for provider API keys (see Provider Integrations)
Then, it looks if the model was previously cached
Finally, it makes a simple request for each un-cached model to check if these models are available